Medical Informatics andPopulation-Based Medicine:

An Introduction to the Clinical Data Repository

Jason Lyman, MD, MS

Department of Health Evaluation Sciences

University of Virginia School of MedicineJanuary 14, 2003

“The U.S. health care system is in need of fundamental change. Americans ought to be able to count on receiving care that is safe and uses the best scientific knowledge. But there is strong evidence that this is not the case. Health care today harms too frequently, and fails to deliver its potential benefits routinely. As medical science and technology have advanced at a rapid pace, the health care delivery system has foundered. Between the care we have and the care we could have lies not just a gap, but a wide chasm.”

Crossing the Quality Chasm: A New Health System for the 21st Century Institute Of Medicine, National Academy of Sciences, 2001

Six Aims for Improvement

Health care needs to be:• Safe

• Effective

• Patient-centered

• Timely

• Efficient

• Equitable

Applying evidence to healthcare delivery

Using information technology

Aligning financial incentiveswith quality improvement

Preparing the workforce

Are we really that far off?

Aren’t our patients safe?

Don’t we deliver effective care?

Don’t we do so equitably?

Health Care Should Be Safe, …

44,000 – 98,000 people die each year from preventable medical errors (defined as the failure of a planned action to be completed as intended, or the use of the wrong plan to achieve an aim)

Costs of errors approximated at $17 billion to $29 billion per year in hospitals nationwide

To Err Is Human: Building a Safer Health SystemInstitute Of Medicine, National Academy of Sciences, 2000

Effective, …

Treatment Compliance

ACE-inhibitors for CHF 50%AIM 1997;157:1103-1108

Aspirin post-MI 85%NEJM 1999;340:286-292

Beta blockers post-MI 60%NEJM 1999;340:286-292

Early reperfusion in elderly <50%JAMA 1999;282:341-348

How long does it take, on average, for new knowledge generated byHow long does it take, on average, for new knowledge generated bya randomized controlled trial to be incorporated into practice? a randomized controlled trial to be incorporated into practice?

And EquitableMortality

• The rate of suicide in American Indian males is twice as high as young white males

Morbidity

• Incidence rates for specific types of cancer (e.g. prostrate) are as much as 60% in African-American males as white males

Health behaviors

• Smoking in adolescents vary widely by geography – 19% in rural counties vs. 11% in non-rural counties

Preventive medicine

• Women below the poverty line are 27% less likely to undergo screening mammography than their counterparts above the poverty line

Access to care

• 28% of children from families at 1-1.5x the poverty level are uninsured vs. 5% in families at 2x or above.

                                         

Roadmaps for Clinical Practice: Primer on Population-Based MedicinePeters & Elster, AMA, January 2002

So…

What factors affect health disparity?

Why is compliance with proven therapies suboptimal?

Why do have so many errors?

What can be done about it?

One Approach: A Population-Based Perspective

Can be helpful for addressing health disparities

Can be useful for evaluating the quality of health care, including the:• processes of care (e.g. efficiency, evidence-based care)

• and outcomes (e.g. morbidity, mortality, complications)

Complements traditional individual-oriented care

Can be incorporated into clinical practice

Today’s Agenda

What is Population-Based Medicine (PBM)?

The role of PBM

Medical Informatics and PBM

The Clinical Data Repository (CDR)• Introduction

• Demo

The CDR as a PBM resource for the POM course

The Continuum of Clinical Care

“It is the duty of the physician to promote and safeguard the

health of the people.”

World Medical Association Declaration of Helsinki, 1964

“ The health of my patient will be my first consideration.”

Declaration of Geneva of the World Medical Association, 1948

What is Population-Based Medicine (PBM)?

“…medicine that addresses the health care of whole populations rather than that of individual patients. It represents a community-based strategy for disease management and health promotion and places each individual patient within the context of the larger community made up of both sick and healthy people.”

Robert Wood Johnson Foundation

PBM In Practice

Identifying health disparity in your own practice / population of patients

Improving chronic disease management for specific conditions (e.g. asthma, diabetes, heart failure)

Identifying at-risk populations in your own practice

Using Healthy People 2010

Implementing practice guidelines (e.g. treatment of acute myocardial infarction)

Integrating preventive service recommendations (e.g. screening, counseling, immunization)

Sample PBM Resources

Healthy People 2010 (http://www.healthypeople.gov/)Healthy Virginians 2010 (http://www.vdh.state.va.us/hv2010/index.html)CDC National Center for Health Statistics (http://www.cdc.gov/nchs/)Agency for Healthcare Research and Quality (http://www.ahrq.gov/)National Guidelines Clearinghouse (http://www.guidelines.gov)Guide to Clinical Preventive Services (USPSTF) (http://www.ahrq.gov/clinic/prevnew.htm)

A Population-Based Approach

Identify and characterize a population of patients• E.g. all the patients of a specific provider, clinic, or healthcare

institution

Identify the health care problems of highest priority• By prevalence, outcome, utilization, etc.

Adapt office procedures to maximize delivery of appropriate services• What is the problem, and how best can it be addressed?

Assess impact and provide feedback

PBM-Related Information Needs

What kind of information do you need to implement these efforts?

Where does the data come from?

Is the data valid?

How is the data recorded and stored?

How is the information provided in a timely fashion?

Medical Informatics

“..the scientific field that deals with the storage, retrieval, sharing, and optimal use of biomedical information, data, and knowledge for problem solving and decision making.”

Shortliffe, 2001

Medicine is Data-Intensive…

Clinical medicine• History, physical exam

• Laboratory tests

• Imaging modalities

• Monitoring data (OR, ICU)

Population health

Health services research• Clinical trials

• Outcomes research

Size of the biomedical literature databases• MEDLINE contains approx. 11 million citations, with 400,000 new

references added each year

…But the Practice of Medicine Can Be Information-Poor

When did this patient last have a colonoscopy?

What is the prognosis for my patient with end stage renal failure?

Did the female I sent for a rheumatology referral manage to get an appointment? Did she go? What happened?

What did this man’s prostate biopsy show?

Where is the *$#@! chart?!

What is Medical Informatics?

Electronic Medical Records (EMR)

Physician Order Entry (POE)

Information Retrieval (IR)

Computerized Decision Support Systems (DSS)

Medical Terminology

Medicine and the Internet

Handheld Computers in Clinical Practice

Multimedia Medical Education

Privacy and Security

A National Perspective on Informatics

“Information technology, including the Internet, holds enormous potential for transforming the health care delivery system, which today remains relatively untouched by the revolution that has swept nearly every other aspect of society.…. the committee calls for a nationwide commitment of all stakeholders to building an information infrastructure to support health care delivery, consumer health, quality measurement and improvement, public accountability, clinical and health services research, and clinical education. This commitment should lead to the elimination of most handwritten clinical data by the end of the decade.” (emphasis added)

Crossing the Quality Chasm: A New Health System for the 21st Century Institute Of Medicine, National Academy of Sciences, 2001

Informatics and PBM

Where do we find all the data we need to incorporate a population-based medicine approach?The holy grail of medical informatics is the development and widespread adoption of comprehensive electronic medical records that collect detailed, longitudinal data about patient encounters• CC, HPI, PMH, FH, SH, Allergies, Meds, Immunizations• Diagnoses (admitting, working, discharge)• Procedures (therapeutic and diagnostic)• Labs, Radiology• Images / text / waveform / numeric• Providers

Electronic Medical Record: The Holy Grail

Discharge Summary

Admission Note

Computer-ReadableCoded Data

Automatic, co

mprehensiv

e

accurate an

d reliab

le

Assist ordering

Error prevention

Facilitate access to knowledge

Research

PBM

The potential!

Reality

Chief complaintHistory and

Physical Daily ProgressNotes

VitalsOrders

Nursing NotesOperative Notes

Dietary Notes DischargeSummariesLaboratory

Results

So, In the Meantime…

Clinicians must find ways to use existing electronic data (demographics, financial, limited coded diagnoses and procedures, labs) • E.g. Dr. Joel Shectman, Internal Medicine – compiling laboratory

and visit data to evaluate the care of diabetic patients at Uva and provide feedback to residents

OR

Implement processes to capture additional data• E.g. Dr. Scott Strayer, Family Medicine – capture of tobacco use

data on handheld computers

• Creation of specialized office information systems*

*A Primer on Population-Based Medicine, AMA, 2002

Information Retrieval: Accessing Population Data

State / regional / national data available for some conditions (CDC, AHRQ)

In most academic and community health care institutions, aggregate patient data is unavailable to health care providers and researchers

Pasteur describes work on germ theory

Hippocrates born

First textbook of anatomypublished by Vesalius

A Brief History of Medicine

460 BC

1543 AD

?

1864

1995

The UVa Clinical Data Repository is Born

What is the CDR?

The UVa Clinical Data Repository (CDR) is a unique local resource for integrating clinical data from Uva patients into a single WWW-based database.• Laboratory data

• Billing data

• Registration data

• Virginia Department of Health (VDH)

Purpose of the CDR

Originally developed as a tool for clinical researchers

Enable flexible and rapid retrospective queries of patient data• “Show me all patients admitted with pneumonia in 1999”

Supply data sets for research and medical management applications

Provide insights into opportunities for improving outcomes and containing costs• What therapies tend be associated with better patient outcomes?

Data Contents

Coded diagnoses and procedures for each inpatient or outpatient visitLaboratory resultsMedications (inpatient visits only)DemographicsUtilization and financial dataMortality data

What’s missing?• Symptoms• Exam findings• Outpatient pharmacy• Microbiology (culture results)• ….

Security and Confidentiality

Disguised Patient and MD Identifiers

Authorization Required for Access

All Uses are Tracked and Audited

2nd Level Authorization Required to obtain real identifiers

The Query Process

1. Define a population

2. Submit your query to the database

3. Display standard reports or download data

Define a Population

By diagnosis• Patient visits for which ACUTE MI was diagnosed

By procedure• Patient visits in which SCREENING MAMMOGRAPHY was

performed

• Patients who received a BLOOD TRANSFUSION

By age, race, gender

By date

By unit

Define a Population

By medication• Patients who received ASPIRIN

By lab test• Patients who had a HCT < 25%

By physician, clinic, age, sex, date range, insurance type, ...

Standard Reports – Summary

How many patients / visits were there that met your criteria?

Demographics

Comorbidities (the presence of other diseases in this patient population)

Diagnoses / procedures that were coded during these visits

Lab results

Mortality

…

Diagnoses and Procedures in the CDR

For every hospitalization, clinic visit, ER visit, etc., the diagnoses assigned and procedures performed must be recorded for the clinic/provider/hospital to be reimbursed!In clinic – done by providers (you’ll get to know this all too well…)In the hospital, trained coders carefully review the chart and make a list of diagnoses assigned.Every visit has one PRINCIPAL diagnosis, but potentially many more SECONDARY diagnoses.

• A patient might be admitted for an asthma exacerbation, but then develop pneumonia, develop a drug allergy, etc.

• Her principal diagnosis would be asthma, and secondary diagnoses would include pneumonia, allergic reaction, etc.

Diagnoses are represented using a system of codes called ICD9.Procedures are represented using CPT (outpatient) and ICD9 (inpatient).

Diagnoses and Procedures

To specify the diagnosis and/or procedures you want to use as conditions, you must identify the ICD-9/CPT codes that match the concepts you’re interested in.

The CDR allows you to search for specific codes or browse through a hierarchical classification (CCHPR).

See CDR FAQ (in the Documentation link of the menu)

More on this later…..

CDR Demonstration

Goals/objectives of Healthy Virginians 2002 include:• Eliminating health disparities• Reducing the hospitalization rate for pediatric asthma (among

other conditions)• Reducing emergency department visits for asthma

Before we can make progress, we need to know where we stand!Let’s take a closer look at pediatric asthma at UVa….• What disparities exist?• How often are kids hospitalized with asthma?• How often do they go to the ED for this condition?

Asthma Prevalence by Race, GenderCount of Ptid RaceHx_of_RAD A B H I O W (blank) Grand TotalN 96.92% 86.76% 94.44% 100.00% 96.37% 93.37% 100.00% 92.03%Y 3.08% 13.24% 5.56% 0.00% 3.63% 6.63% 0.00% 7.97%Grand Total 100.00% 100.00% 100.00% 100.00% 100.00% 100.00% 100.00% 100.00%

Count of Ptid SexHx_of_RAD F M Grand TotalN 93.75% 90.48% 92.03%Y 6.25% 9.52% 7.97%Grand Total 100.00% 100.00% 100.00%

Insurance, Utilization, and Asthma

  Percentage with a history of

asthma

Average Number of ER Visits (over

a five-year period)

Self-Pay 9.19 1.51

Medicaid 9.35 1.23

Other 6.55 .57

ED Visits for Asthma  In those patients with a history of RAD, what percentage have had the

following number of ER visits for RAD: 

None 1-3 4-6 7 or more

Self-Pay 69.75 28.71 0.96 0.58

Medicaid 75.40 22.24 1.60 0.75

Other 83.11 14.57 1.77 0.54

Our Goals for PBM / CDR in POM

Introduce you to the principles and promise of population-based medicine

Provide you with access to aggregate data to explore these principles first hand

Teach you how to use the CDR

CDR and POM

Three PBM exercises throughout the course of the semester• Two students per group will complete a given exercise in a special

session held in the library classroom (Carter). Each student will attend one of these sessions.

The exercises will focus on specific PBM objectives as described in this lecture, for example• What heath disparities exist in UVa patients?• How can our data be used to measure quality of care?• How are our patients with chronic illness being managed and how

might we improve?• How well are preventive services being administered?

How else might you use the CDR?

A helpful tool for comparing information in textbooks with real patient data (disease prevalence and practice patterns can vary widely) – e.g. how do physicians here treat meningitisComing into a new rotation:• what kinds of patients will I see? • What kinds of conditions?

Learning about patients with complex chronic illnesses• What diseases / complications is my patient with chronic renal

failure at risk for?

Potential source of data for a research project

Need Help?

On-line code tables and documentation

Available CDR project team-members• Jason Lyman, MD, MS, CDR Director

(lyman@virginia.edu, 924-8240)

• Ken Scully, MS, Developer and Database Administrator

(kws9s@virginia.edu, 982-4035)

Next Steps

We will let you know the dates that the PBM sessions are scheduled

Stay tuned…

Please fill out the CDR Access Request Form (top half only) and return to me or Robin Stevens